Plasma testosterone levels and aggressive behavior were measured in 12 women with and without perimenstrual affective symptomatology (e.g., depression, irritability) during the menstrual, midfollicular, ovulatory, and premenstrual phases of the menstrual cycle. The Point Subtraction Aggression Paradigm was used to quantify aggressive response to provocation. Subjects had two response options: a point-maintained option (100 presses earned a point worth 10 cents) and an aggressive response option (10 presses ostensibly subtracted a point from a fictitious partner's counter). Subjects were provoked by the periodic subtraction of a point that was attributed to the responding of a fictitious opponent. Although plasma testosterone levels (determined by radioimmunoassay) increased significantly during the ovulatory phase, aggressive response to provocation remained unchanged across the menstrual cycle. Plasma testosterone did not differ between the 2 groups during any phase. A relationship between plasma testosterone levels and use of the aggressive response option was seen only during the midfollicular phase (Spearman r = .673, p = .017). These preliminary data suggest that: 1. The relationship in female subjects between endogenous testosterone and aggressive behavior is inconsistent; 2. self-report of perimenstrual symptomatology is a more consistent predictor of aggressive behavior across the menstrual cycle than plasma testosterone; and 3. perimenstrual emotional symptomatology is not related to testosterone levels.